Introduction

Obesity is a growing pandemic with a huge burden of associated harmful effects on both physical and mental health. The role of bariatric surgery is well established in reducing morbidity from obesity. Bariatric surgery results in greater weight loss and greater improvement in weight-associated comorbidities compared with non-surgical interventions, regardless of the type of procedures used [1].

Overweight and obese women are at risk of urinary storage symptoms and lower urinary tract symptoms including almost double the risk of stress urinary incontinence (SUI) and a three times increased risk of urge urinary incontinence [2, 3]. Previous systematic reviews have addressed different aspects of the relationship between obesity and treatments for obesity and incontinence [4]. Non-surgical weight loss interventions are known to be associated with an improvement in urinary incontinence (UI) in overweight and obese women [2, 5]. Current incontinence guidance from bodies including The National Institute for Health and Care Excellence (NICE) [6] and the European Association of Urology (EAU) [7] recommends weight loss for obese women, as non-surgical weight loss through a variety of lifestyle and dietary interventions has been consistently associated with benefits for UI [3, 8, 9] even for women with moderately increased BMI.

Multiple cohort studies have tested the effects on UI of surgical weight loss for the morbidly obese. One earlier systematic review reported pooled benefits for specific incontinence scales, but was able to include only a minority of available studies, and could not differentiate between subtypes of UI [10].

Our aim was to systematically assess all available studies reporting the effect of bariatric surgery on changes in BMI in women with urinary incontinence. We planned to quantify the pooled effect sizes on quality of life (QoL) and cure, and test for associations between effects on weight loss and effect on UI across different procedures.

Materials and methods

We followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidance [11]. No ethical approval was required.

Eligibility criteria

Studies were eligible if they were randomised controlled trials, case–control and cohort studies (either retrospective or prospective); if they included obese and overweight individuals who have undergone bariatric surgery; if they had used validated urinary or pelvic floor questionnaires; and if they had a follow-up of at least 6 months.

Search strategies

We identified relevant studies using the keywords “urinary incontinence,” “bariatric surgery,” “surgically induced weight loss,” and obesity. We searched MEDLINE, Embase and the Cochrane library up to September 2016. The references of all eligible studies were also hand-searched. We also searched abstracts published at the annual meetings of the American Urological Association (AUA), European Association of Urology (EAU), International Continence Society (ICS) and International Urogynecological Association (IUGA) up to September 2016. Our search was restricted to humans and the English language.

Two independent reviewers (BP, GC) independently screened titles and abstracts first, and subsequently full texts, with duplicate data extraction. Disagreements were resolved by consensus.

Data collection and risk of bias assessment

No randomised controlled trials were identified. All identified studies were cohort studies. Two independent reviewers categorised studies as being either at a low or at a high risk of bias using a novel instrument developed by CLARITY group specifically designed for longitudinal symptom research studies aimed at the general population (Tables 1, 2), evaluating the representativeness of the source populations, the accuracy of the outcome assessment and the proportion of missing data [31].

Table 1 General characteristics of studies included in the systematic review
Full size table
Table 2 More general characteristics of studies included in the systematic review
Full size table

Data synthesis and statistical analysis

We categorised the method of diagnosis of urinary incontinence: self-report, structured clinical interview, objective testing (cough test/pad test), or patient completed questionnaires; and the criterion for diagnosis, i.e. the number of episodes weekly/monthly. Included studies used different validated questionnaires with some studies using more than two validated questionnaires. Where a single study provided multiple estimates of cure or improvement, we selected the most widely employed or validated measure for pooling. Symptom and QoL scores were standardised on a scale of 0–100, whereas measures of incontinence cure were converted to proportions to enable pooling.

Data were analysed using metan in Stata 14, using the metafunnel, metabias and metareg packages. The Metaprop [32] command was used to pool cure rates, as standard weighting methods produce inaccurate 95% confidence intervals with binomial data, where rates approach 0% or 100%. The lower and upper confidence intervals were computed using the “exact” or Clopper–Pearson method. Changes in symptom scores and QoL were pooled with inverse-variance weights obtained from a random-effects model. We calculated heterogeneity between studies with Higgins’ I2. We explored heterogeneity using meta-regression, testing the type of bariatric surgery and change in BMI as predictors of effect size.

Results

Literature search and study characteristics

Thirty full-text articles were screened after reviewing abstracts and titles, and 23 full-text articles were included in the meta-analysis (Fig. 1).

Fig. 1
figure 1

Four-phase flow diagram of the systematic reviewing process

Full size image

From a total of 23 studies (n = 3,225), 17 studies (n = 1,069) provided data for the change in the urinary scores [13,14,15,16,17, 19,20,21,22,23,24,25,26, 28,29,30, 33] and 17 studies (n = 1506) studies provided data for the proportion of women cured of any UI [12,13,14,15,16,17,18,19,20,21, 24,25,26,27,28, 30, 34] and 8 studies (n = 377) provided data for the proportion of women cured of SUI [12, 13, 21, 24, 30, 33,34,35]. We could not collate data for UUI because of inconsistencies in data reporting.

Twenty studies specified the type of bariatric procedure, whereas three studies did not give information. Gastric bypass was performed exclusively in 6 studies, whereas gastric banding was carried out in 2 studies, and 12 studies included women with different types of bariatric surgery, including sleeve gastrectomy, and diversion surgery. Four studies had follow-up data of 6 months, nine studies had a follow-up of 12 months and 10 studies had a follow-up of more than 1 year (maximum of up to 44 months).

Data regarding BMI was available for 21 studies. Pre-operative and post-operative BMI data were available to assess the change in BMI in 15 studies. The general characteristics of individual studies are presented in Tables 1 and 2 and urinary scores data are shown in Table 3.

Table 3 Outcomes (urinary scores) of the studies
Full size table

Risk of bias

Of the 23 studies included, 17 studies (74%) had a high risk of bias and 6 (26%) had a low risk of bias (Table 4). Of these 23 studies, 17 (74%) had data both at baseline and at follow-up; 10 (43%) had few missing data in the follow-up, and 6 (17%) used representative source populations.

Table 4 Risk of bias of included studies
Full size table

Changes in BMI

Changes in BMI were pooled for 15 studies. Bariatric surgery was associated with a significant reduction in BMI (13%, 95% confidence interval = −10.797 to -15.619, p < 0.001), but with high heterogeneity, I2 = 94.9%, p = 0.0001 (Fig. 2).

Fig. 2
figure 2

Forest plot displaying the mean change in body mass index after bariatric surgery. ES effect size, CI confidence interval

Full size image

Change in urinary scores

The changes in urinary scores were pooled for 17 studies. A subgroup analysis based on the type of surgery for studies showed 14% improvement in the urinary scores of patients after bariatric surgery (weighted mean difference = −14.79; CI = −18.47 to −11.11), with substantial heterogeneity (I2 = 87.1%), as shown in Fig. 3.

Fig. 3
figure 3

Forest plot displaying the mean change in urinary scores after bariatric surgery based on the type of surgery. WMD weighted mean difference

Full size image

Results were unchanged in sensitivity analyses excluding each study once. We explored the heterogeneity using meta-regression, testing the type of bariatric surgery and change in BMI as predictors of effect size, and none of them was associated with the change in urinary scores.

Proportion of women cured

Changes in proportions of women cured of any UI were pooled for 17 studies. For overall UI, the pooled cure rate was 59% (95% CI = 51 to 66), again with high heterogeneity (I2 = 84.93%, p < 0.0001 (Fig. 4). Changes in proportions of women cured of SUI were pooled for 8 studies. The pooled cure rate of any SUI was 55% (95% CI = 40 to 70) with high heterogeneity (I2 = 89.91%, p < 0.0001; Fig. 5).

Fig. 4
figure 4

Meta-analysis of proportions of women cured of any urinary incontinence (UI) after bariatric surgery (Metaprop)

Full size image
Fig. 5
figure 5

Meta-analysis of proportions of women cured of stress urinary incontinence (SUI) after bariatric surgery (Metaprop)

Full size image

A subgroup analysis based on the type of surgery did not show any significant differences in the proportion of women cured across procedures (Fig. 4).

Discussion

Strengths

To our knowledge, this is the first systematic review of the effect of weight loss surgery on urinary incontinence reporting cure rates. We used pre-specified criteria for the inclusion of studies, considered grey literature and carefully avoided duplicate data. We used appropriate statistical methods, and further sensitivity analysis did not change the results. Our study quantitatively summarised the available evidence for the effect of weight loss after bariatric surgery on urinary incontinence.

Limitations

There were 17 studies, which reported standardised quality of life urinary scores, but most of the studies used different quality of life scores and heterogeneity in the meta-analysis is high. Second, there were few studies with small numbers, which can skew the true effect size. The smallest included study consisted of 12 patients. Although the differences in weight loss led to the improvement in urinary scores and the proportion of women cured, on sensitivity analysis, we did not have the power to show that (using study-level statistics) the magnitude of weight loss is associated with improvement in UI. If we had individual patient data, we would have been able to explore in greater depth the change in BMI and improvement in urinary incontinence.

Another important limitation is the variable follow-up period (6–44 months), which influences the proportion cured of any UI and SUI. Also, there were insufficient data to report cure rates for UUI.

Though most studies defined their inclusion and exclusion criteria, none of the studies commented on co-interventions like conservative measures, pelvic floor exercises or medical treatment during the study period. Apart from three studies in which controls were matched, none of the other studies had a control group. This may indicate that the other treatments/factors apart from bariatric surgery may have influenced the results.

The eight studies that provided data for the cure rates of SUI had not commented on interventions such as pelvic floor exercises and this could have an impact on published data for cure rates of SUI.

The other limitations include response bias, as sometimes standardised questionnaires were filled out during telephone/clinic interviews where patients may have answered favourably towards what researchers want to hear, and there was lack of blinding in all the studies. The patients in these trials were recruited while women were on the waiting list for bariatric surgery and questionnaires were distributed to identify suitable women for the study. These studies relied on self-reported data, which is influenced by recall bias.

Implications for clinical practice and future research

This systematic review and meta-analysis showed improvement in the quality of life scores of women with urinary incontinence with a significant proportion of women achieving cure in any UI and SUI. These results make bariatric surgery a promising treatment option for obese and overweight women in whom conservative/pharmacological methods of losing weight have failed. In our SR, the degree of change in BMI was not the predictor of effect size. A subgroup analysis based on different types of surgery showed that greater weight loss (mean BMI reduction by 14%) was achieved with gastric bypass surgery compared with gastric banding (mean BMI reduction by 8%), but there was no difference in improvement in UI scores between the two subgroups. a higher degree of improvement in urinary symptom scores.

The remaining women with UI demonstrated some improvement, no change or worsening of symptoms, although they all achieved some degree of surgical weight loss. The data for these women could not be analysed because of inconsistencies in reporting in the articles. This suggests that women who achieved weight loss might have showed a trend towards improvement, although they may not have been completely cured of UI, especially UUI.

Large prospective individual studies have shown that individuals regained weight with time and associated with this is the relapse of co-morbidities [36]. Although bariatric surgery has a more sustained effect on weight loss compared with non-surgical methods of weight loss, it is associated with serious complications and a risk of death [37].

These results will help us to counsel women that weight loss surgery is beneficial in achieving improvement in urinary incontinence. We know from the EpiLUTS study that women are far more likely to report any UI, and, in particular, SUI, and SUI may be particularly sensitive to increased weight above the normal range in women [38]. When conservative measures fail to treat UI and surgical interventions are being considered, it seems appropriate to consider bariatric surgery in overweight and obese women in addition to considering specific surgery for urinary incontinence. Given serious complications associated with bariatric surgery and the lack of long-term follow up, careful consideration of other options and detailed counselling are needed before offering this surgery for the improvement of the quality of life of women with urinary incontinence.

Conclusions

Evidence from these cohort studies suggests that there might be a clinically meaningful improvement in urinary symptom scores, the proportion of women cured of SUI and any UI, and a reduction in BMI after bariatric surgery, but with substantial differences between studies.

There is a lack of data regarding improvement/cure in women with UUI, and future studies should report such data to guide our management of these women.

Current data are limited to short-term follow-up of these women, with great heterogeneity in the studies. Further studies are needed to explore the impact of bariatric surgery on individual subtypes of incontinence, and to investigate which types of surgery offer the largest benefit for LUTS.

This review will help patients and surgeons to counsel overweight and obese women for weight loss surgery as one of the treatment options for urinary incontinence.